h. ray, university of florida1 miniboone e e
TRANSCRIPT
H. Ray , University of Florida 1
MINIBOONE
e
e
H. Ray , University of Florida 2
The Notorious LSND Result
• E =20-52.8 MeV• L =25-35 meters• 3.8 excess• Different from other
oscillation signals• Higher m2
• Smaller mixing angle• Much smaller
probability (very small signal) ~0.3%
e
Posc =sin22 sin2 1.27 m2 L
E
H. Ray , University of Florida 3
Testing LSND
• Want same L/E• Different detection
method• Different sources of
systematic errors• Test for oscillations
using neutrinos, anti-neutrinos
H. Ray , University of Florida 4
MiniBooNE Neutrino Beam
• Mesons decay into the anti-neutrino beam seen by the detector
K- / - - +
- e- + + e
• MiniBooNE L/E = 0.5 km / 0.8 GeV (.625)• LSND L/E = 0.03 km / 0.05 GeV (.6)• Anti-nu analysis uses 3.386E20 protons on target
H. Ray , University of Florida 5
MiniBooNE Detector
• 12.2 meter diameter sphere• Pure mineral oil• 2 regions
Inner light-tight region, 1280 PMTs (10% coverage) Optically isolated outer veto-region, 240 PMTs
H. Ray , University of Florida 6
Event Signaturecharged-current quasi-elastic events νe p → e+ n
H. Ray , University of Florida 7
MiniBooNE Analysis
• MiniBooNE is looking for e oscillations
• MiniBooNE is looking for an excess of e
(appearance)
• MiniBooNE performing a blind analysis Some of the info in all of the data
Charge per PMT as a function of time All of the info in some of the data
~oscillation free sample to be open for analysis Low-E NC elastic events, CCQE, CC+
e candidate events locked away until the analysis 100% complete
Same strategy as neutrino mode
H. Ray , University of Florida 8
Neutrino Mode Result• Used 2 complementary
analyses• ruled out interpretation
of LSND as νμ→νe oscillations two-nu oscillations standard L/E depend. no CP, CPT violation
• Unexplained excess of events in low-energy region Excess incompat. With
2-nu lsnd-style oscillations
Phys. Rev. Lett. 98, 231801 (2007)
H. Ray , University of Florida 9
Anti-neutrino Analysis• Similarities with neutrino mode
Using two complementary analysesUsing same algorithmsSame event selection cuts
• Changes made wrt neutrino modeAdded an event selection cut that removes
most of the dirt (out of tank) background at low energy
Added k- production systematic error New NC π0 measurementNew dirt fraction extracted
H. Ray , University of Florida 10
Analysis Chain
Beam Flux Prediction
Cross Section Model
Optical ModelEvent
ReconstructionParticle
Identification
Simultaneous Fit to anti-
and anti- e events
Start with a Geant 4 flux prediction for the nu spectrum from and K produced at the target
Use track-based event reco
Predict nu interactions using the Nuance event generator
Pass final state particles to Geant 3 to model particle, light propagation in the tank
Fit reco. energy spectrum for oscillations
Use hit topology and timing to identify electron-like or muon-like charged current neutrino interactions
Track Based Analysis
H. Ray , University of Florida 11
Analysis Chain
Beam Flux Prediction
Cross Section Model
Optical ModelEvent
ReconstructionParticle
Identification
Simultaneous Fit to anti-
and anti- e events
Start with a Geant 4 flux prediction for the nu spectrum from and K produced at the target
Use point-source event reco
Predict nu interactions using the Nuance event generator
Pass final state particles to Geant 3 to model particle, light propagation in the tank
Fit reco. energy spectrum for oscillations
Use boosted decision tree to identify electron-like or muon-like charged current neutrino interactions
Boosted Decision tree
H. Ray , University of Florida 12
Expected Backgrounds
Boosting Analysis Track Based Analysis
500 MeV < Reconstructed Neutrino Energy < 1.25 GeV
Intrinsic e backgrounds mis-id backgrounds
3.386e20 POT
H. Ray , University of Florida 13
Expected Sensitivity
As in neutrino mode, TBA is the more sensitive (default) analysis.
BDT is the cross-check analysis
H. Ray , University of Florida 14
Expected Background EventsBackground composition for TBA analysis (3.386e20 POT)
LSND best-fit (Δm2, sin22θ)=(1.2,0.003)
[note: statistical-only errors shown]e from ±, ± 8.44 17.14
e from k±,0 8.20 24.88
Other e 1.11 1.21
CCQE 2.86 1.24
NC 0 24.60
7.17
Delta Rad 6.58 2.02
Dirt 4.69 1.92
Other 3.82 2.20
Total Bgd 60.29
57.78
LSND Best Fit 4.33 12.63
Intr
insi
c e
m
is-i
d
200-475 MeV
475-1250 MeV
H. Ray , University of Florida 15
Systematic Uncertainties
Source
EνQE range (MeV) 200-475 475-
1100200-475 475-
1100
Flux from π+/μ+ decay
0.4 0.7 1.8 2.2
Flux from π-/μ- decay
3.3 2.2 0.1 0.2
Flux from K+ decay 2.3 4.9 1.4 5.7
Flux from K- decay 0.5 1.1 - -
Flux from K0 decay 1.5 5.7 0.5 1.5Target and beam models 1.9 3.0 1.3 2.5
ν-cross section 6.4 12.9 5.9 11.9NC π0 yield 1.7 1.6 1.4 1.9
Hadronic interactions 0.5 0.6 0.8 0.3
External interactions (dirt) 2.4 1.2 0.8 0.4
Optical model 9.8 2.8 8.9 2.3
Electronics & DAQ model 9.7 3.0 5.0 1.7
Total (unconstrained) 16.3 16.2 12.3 14.2
From fits to world’s data (HARP, Kaon production)
Track Based AnalysisAntinu Mode Nu Mode
H. Ray , University of Florida 16
Systematic Uncertainties
Source
EνQE range (MeV) 200-475 475-
1100200-475 475-
1100
Flux from π+/μ+ decay 0.4 0.7 1.8 2.2
Flux from π-/μ- decay 3.3 2.2 0.1 0.2
Flux from K+ decay 2.3 4.9 1.4 5.7
Flux from K- decay 0.5 1.1 - -
Flux from K0 decay 1.5 5.7 0.5 1.5
Target and beam models 1.9 3.0 1.3 2.5
ν-cross section 6.4 12.9 5.9 11.9
NC π0 yield 1.7 1.6 1.4 1.9Hadronic interactions 0.5 0.6 0.8 0.3
External interactions (dirt)
2.4 1.2 0.8 0.4
Optical model 9.8 2.8 8.9 2.3
Electronics & DAQ model 9.7 3.0 5.0 1.7
Total (unconstrained) 16.3 16.2 12.3 14.2
Track Based Analysis
Internal MB measurements
Antinu Mode Nu Mode
H. Ray , University of Florida 17
Systematic Uncertainties
Source
EνQE range (MeV) 200-475 475-
1100200-475 475-
1100
Flux from π+/μ+ decay 0.4 0.7 1.8 2.2
Flux from π-/μ- decay 3.3 2.2 0.1 0.2
Flux from K+ decay 2.3 4.9 1.4 5.7
Flux from K- decay 0.5 1.1 - -
Flux from K0 decay 1.5 5.7 0.5 1.5
Target and beam models
1.9 3.0 1.3 2.5
ν-cross section 6.4 12.9 5.9 11.9
NC π0 yield 1.7 1.6 1.4 1.9
Hadronic interactions 0.5 0.6 0.8 0.3
External interactions (dirt) 2.4 1.2 0.8 0.4
Optical model 9.8 2.8 8.9 2.3
Electronics & DAQ model
9.7 3.0 5.0 1.7
Total (unconstrained) 16.3 16.2 12.3 14.2
Track Based AnalysisAntinu Mode Nu Mode
Determined by special runs of the beam MC. All beam uncertainties not coming from meson production by 8 GeV protons are varied one at a time. variations are treated as 1σ excursions, prop into a final error matrix.
Uncert. in a number of hadronic processes, mainly photonuclear interaction final state
uncertainties in light creation, propagation, and detection in the Tank. Use set of 130 MC “multisims” that have been run where all these parameters are varied according to their input uncertainties.
H. Ray , University of Florida 18
Systematic Uncertainties
Source
EνQE range (MeV) 200-475 475-
1100200-475 475-
1100
Flux from π+/μ+ decay 0.4 0.7 1.8 2.2
Flux from π-/μ- decay 3.3 2.2 0.1 0.2
Flux from K+ decay 2.3 4.9 1.4 5.7
Flux from K- decay 0.5 1.1 - -
Flux from K0 decay 1.5 5.7 0.5 1.5
Target and beam models 1.9 3.0 1.3 2.5
ν-cross section 6.4 12.9 5.9 11.9
NC π0 yield 1.7 1.6 1.4 1.9
Hadronic interactions 0.5 0.6 0.8 0.3
External interactions (dirt) 2.4 1.2 0.8 0.4
Optical model 9.8 2.8 8.9 2.3
Electronics & DAQ model 9.7 3.0 5.0 1.7
Total (unconstrained) 16.3 16.2 12.3 14.2
Track Based Analysis
Fit to the anti-νμ CCQE and anti-νe candidate spectra simultaneously. takes advantage of strong correlations between anti-νe signal, background, and the anti-νμ CCQE sample, reduces systematic uncertainties and constrains intrinsic anti-νe from muon decay
Antinu Mode Nu Mode
Antineutrino appearance search is statistics limited!
H. Ray , University of Florida 19
Counting Expt :475 MeV < E
QE< 1.250 GeV61 observed evts57.8 ± 10.0 expected eventsExcess over background : 0.3
Counting Expt :200 MeV < E
QE< 475 MeV61 observed evts61.5 ± 11.7 expected eventsExcess over background : -0.04
Observed Event Distribution
H. Ray , University of Florida 20
Excess Distribution
(Δm2, sin22θ) = (4.4 eV2, 0.004)
χ2best-fit(dof) = 18.18 (17)
χ2-probability = 37.8%
H. Ray , University of Florida 21
EνQE range (MeV) anti-ν mode ν mode
(3.386e20 POT) (6.486e20 POT)
200-300
300-475
200-475
475-1250
Excess
Deficit
Data 24 232MC ± sys+stat (constr.) 27.2 ± 7.4 186.8 ± 26.0Excess (σ) -3.2 ± 7.4 (-0.4σ) 45.2 ± 26.0 (1.7σ)
Data 37 312MC ± sys+stat (constr.) 34.3 ± 7.3 228.3 ± 24.5Excess (σ) 2.7 ± 7.3 (0.4σ) 83.7 ± 24.5 (3.4σ)
Data 61 544MC ± sys+stat (constr.) 61.5 ± 11.7 415.2 ± 43.4Excess (σ) -0.5 ± 11.7 (-0.04σ) 128.8 ± 43.4 (3.0σ)
Data 61 408MC ± sys+stat (constr.) 57.8 ± 10.0 385.9 ± 35.7Excess (σ) 3.2 ± 10.0 (0.3σ) 22.1 ± 35.7 (0.6σ)
arXiv:0812.2243
Event Summary
H. Ray , University of Florida 22
Same ν and ν NC cross-section (HHH axial anomaly), POT scaled, Low-E Kaon scaled: strongly disfavored as an explanation of the MiniBooNE
low energy excess!
most preferred modeL: low-energy excess comes from neutrinos in the beam (no contribution from anti-neutrinos)
Currently in process of more careful consideration of correlation of systematics in neutrino and antineutrino mode… results coming
soon!
Stat Only Correlated Syst Uncorrelated Syst
Same ν,ν NC 0.1% 0.1% 6.7%NC π0 scaled 3.6% 6.4% 21.5%POT scaled 0.0% 0.0% 1.8%Bkgd scaled 2.7% 4.7% 19.2%CC scaled 2.9% 5.2% 19.9%Low-E Kaons 0.1% 0.1% 5.9%ν scaled 38.4% 51.4% 58.0%
Maximum χ2 probability from fits to ν and ν excesses in 200-475 MeV range
Preliminary
(upper and lower bounds)
_
_
Implications for Low-E Excess
H. Ray , University of Florida 23
Final Limits
• No strong evidence for oscillations in anti-nu mode Analysis limited by stat
• No evidence of excess at low energy in anti-nu mode Combine nu, anti-nu for
low-E analysis
• Data collection continuing through June, 09
• NuMI analysis in 2009!
3.386e20 POT
Backup Slides
H. Ray , University of Florida 25
The Notorious LSND
• 800 MeV proton beam + H20 target, Copper beam stop
• 167 ton tank, liquid scintillator, 25% PMT coverage
• E =20-52.8 MeV
• L =25-35 meters
• e + p e+ + n n + p d + (2.2 MeV)
e
H. Ray , University of Florida 26
Neutrino Event Composition
Boosting Analysis
Other12%
NCpi018%
Rad. Delta4%
K+26%
Ko6%
Pi+1%
Mu+33%
Likelihood Analysis
intrinsic e
mis-id
H. Ray , University of Florida 27
Flux Prediction
• Much larger wrong sign component in anti-neutrino analysis
nu mode Anti-nu mode
Intrinsic background to oscillation
analysis
Intrinsic background to oscillation
analysis
Wrong sign
contamination Wrong sign
contamination
H. Ray , University of Florida 28
Systematic Errors• Use Multisims to calculate
systematic errors• In each MC event vary all
parameters at once, according to a full covariance matrix Ex : Feynman scaling of K+
varies 8 params simultaneously
• Vary full set of parameters many times per event
• OM = 70 multisims. All other errors = 1000
• Use this information to form an error matrix
Central Value MC
#evts in 1 E bin
# m
ult
isim
s K+ error
H. Ray , University of Florida 29
Systematic ErrorsError Track vs
Boosting (%)
Checked or Constrained by
data
DAQ 7.5 / 10.8
Target/Horn 2.8 / 1.3
+ Flux 6.2 / 4.3
K+ Flux 3.3 / 1.0
K0 Flux 1.5 / 0.4
Dirt 0.8 / 3.4
NC 0 yield 1.8 / 1.5
Neutrino Xsec 12.3 / 10.5
OM 6.1 / 10.5
H. Ray , University of Florida 30
H. Ray , University of Florida 31
Expected Event Composition
W+
p
n
μ+ can……capture on C…decay too quickly…have too low energy
Nevents 200-475 MeV 475-1250 MeV
intrinsic νe 17.7443.23from π±/μ± 8.44 17.14from K±, K0 8.20 24.88other νe 1.11 1.21
mis-id νμ 42.54 14.55CCQE 2.86 1.24NC π0 24.60 7.17Δ radiative 6.58 2.02Dirt 4.69 1.92other νμ 3.82 2.20
Total bkgd 60.29 57.78
LSND best fit 4.33 12.63
H. Ray , University of Florida 32
Expected Event Composition
Z
A
Nevents 200-475 MeV 475-1250 MeV
intrinsic νe 17.7443.23from π±/μ± 8.44 17.14from K±, K0 8.20 24.88other νe 1.11 1.21
mis-id νμ 42.54 14.55CCQE 2.86 1.24NC π0 24.60 7.17Δ radiative 6.58 2.02Dirt 4.69 1.92other νμ 3.82 2.20
Total bkgd 60.29 57.78
LSND best fit 4.33 12.63
Coherent π0 production
A
γ
γ
H. Ray , University of Florida 33
Expected Event Composition
Z
p
Nevents 200-475 MeV 475-1250 MeV
intrinsic νe 17.7443.23from π±/μ± 8.44 17.14from K±, K0 8.20 24.88other νe 1.11 1.21
mis-id νμ 42.54 14.55CCQE 2.86 1.24NC π0 24.60 7.17Δ radiative 6.58 2.02Dirt 4.69 1.92other νμ 3.82 2.20
Total bkgd 60.29 57.78
LSND best fit 4.33 12.63
Resonant π0 production
p
γ
γ
H. Ray , University of Florida 34
Expected Event Composition
Z
p
Nevents 200-475 MeV 475-1250 MeV
intrinsic νe 17.7443.23from π±/μ± 8.44 17.14from K±, K0 8.20 24.88other νe 1.11 1.21
mis-id νμ 42.54 14.55CCQE 2.86 1.24NC π0 24.60 7.17Δ radiative 6.58 2.02Dirt 4.69 1.92other νμ 3.82 2.20
Total bkgd 60.29 57.78
LSND best fit 4.33 12.63
…and some times… Δ radiative decay
p
H. Ray , University of Florida 35
Expected Event Composition
shower
dirt
Nevents 200-475 MeV 475-1250 MeV
intrinsic νe 17.7443.23from π±/μ± 8.44 17.14from K±, K0 8.20 24.88other νe 1.11 1.21
mis-id νμ 42.54 14.55CCQE 2.86 1.24NC π0 24.60 7.17Δ radiative 6.58 2.02Dirt 4.69 1.92other νμ 3.82 2.20
Total bkgd 60.29 57.78
LSND best fit 4.33 12.63
MiniBooNE detector
H. Ray , University of Florida 36
How consistent are excesses in neutrino and antineutrino mode under different underlying hypotheses as the source of the low energy excess in neutrino mode?
Data 61 544MC ± sys+stat (constr.) 61.5 ± 11.7 415.2 ± 43.4Excess (σ) -0.5 ± 11.7 (-0.04σ) 128.8 ± 43.4 (3.0σ)
200-475 MeV νν_
Scales with POT Same NC cross section for neutrinos and antineutrinos Scales as π0 background Scales with neutrinos (not antineutrinos) Scales with background Scales as the rate of Charged-Current interactions Scales with Kaon rate at low energy
H. Ray , University of Florida 37
Data 61 544MC ± sys+stat (constr.) 61.5 ± 11.7 415.2 ± 43.4Excess (σ) -0.5 ± 11.7 (-0.04σ) 128.8 ± 43.4 (3.0σ)
200-475 MeV
• Performed 2-bin χ2 test for each assumption• Calculated χ2 probability assuming 1 dof
The underlying signal for each hypothesis was allowed to vary (thus accounting for the possibility that the observed signal in neutrino mode was a fluctuation up, and the observed signal in antineutrino mode was a fluctuation down), and an absolute χ2 minimum was found.
• Three extreme fit scenarios were considered: Statistical-only uncertainties Statistical + fully-correlated systematics Statistical + fully-uncorrelated systematics
νν_
H. Ray , University of Florida 38
Eg.: scales with POT (e.g. “paraphotons”,…)
Antineutrino POT: 3.386e20 Antineutrino POTNeutrino POT: 6.486e20 Neutrino POT
One would expect a ν excess of ~(128.8 events)*0.52 = ~67 events
Data 61 544MC ± sys+stat (constr.) 61.5 ± 11.7 415.2 ± 43.4Excess (σ) -0.5 ± 11.7 (-0.04σ) 128.8 ± 43.4 (3.0σ)
200-475 MeV
Obviously this should be highly disfavored by the data, but one could imagine a scenario where the neutrino mode observed excess is a fluctuation up from true underlying signal and the antineutrino mode excess is a fluctuation down, yielding a lower χ2…
= 0.52
νν_
H. Ray , University of Florida 39
Eg.: Same NC cross section for neutrinos and antineutrinos (e.g., HHH axial anomaly)
Expected rates obtained by integrating flux across all energies for neutrino mode, and antineutrino mode
Data 61 544MC ± sys+stat (constr.) 61.5 ± 11.7 415.2 ± 43.4Excess (σ) -0.5 ± 11.7 (-0.04σ) 128.8 ± 43.4 (3.0σ)
200-475 MeV νν_
[Harvey, Hill, and Hill, hep-ph0708.1281]
H. Ray , University of Florida 40
Eg.: Scales as π0 background (same NC ν and ν cross-section ratio)
Expected rates obtained by integrating flux across all energies for neutrino mode, and antineutrino mode
Mis-estimation of π0 background? Or other Neutral-Current process?
For π0 background to fully account for MB ν mode excess, it would have to be mis-estimated by a factor of two……but we have measured MB π0 event rate to a few percent!
Data 61 544MC ± sys+stat (constr.) 61.5 ± 11.7 415.2 ± 43.4Excess (σ) -0.5 ± 11.7 (-0.04σ) 128.8 ± 43.4 (3.0σ)
200-475 MeV νν_
[Phys. Lett. B664, 41 (2008)]
_
H. Ray , University of Florida 41
Eg.: Scales with neutrinos (in both running modes)
In neutrino mode, 94% of flux consists of neutrinosIn antineutrino mode, 82% of flux consists of antineutrinos, 18% of flux consists of neutrinos
Predictions are allowed to scale according to neutrino content of the beam
Data 61 544MC ± sys+stat (constr.) 61.5 ± 11.7 415.2 ± 43.4Excess (σ) -0.5 ± 11.7 (-0.04σ) 128.8 ± 43.4 (3.0σ)
200-475 MeV νν_
H. Ray , University of Florida 42
χ2null(dof)1
χ2best-fit(dof)1 χ2
LSND best-
fit(dof)χ2-prob χ2-prob χ2-prob
20.18(19) 18.18(17) 20.14(19)38.4% 37.8% 38.6%
17.88(16) 15.91(14) 17.63(16)33.1% 31.9% 34.6%
EνQE fit
> 200 MeV
> 475 MeV
EνQE > 200 MeV and Eν
QE > 475 MeV fits are consistent with each other.No strong evidence for oscillations in antineutrino mode.
(3.386e20 POT)
(1Covariance matrix approximated to be the same everywhere by its value at best fit point)
H. Ray , University of Florida 43
2D global fit